Association between clinical frailty, illness severity and post-discharge survival: a prospective cohort study of older medical inpatients in Norway

Abstract

Purpose: Study associations between frailty, illness severity and post-discharge survival in older adults admitted to medical wards with acute clinical conditions.

Methods: Prospective cohort study of 195 individuals (mean age 86; 63% females) admitted to two medical wards with acute illness, followed up for all-cause mortality for 20 months after discharge. Ward physicians screened for frailty and quantified its degree from one to eight using Clinical Frailty Scale (CFS), while clinical illness severity was estimated by New Early Warning Score 2 (NEWS2) and laboratory illness severity was calculated by a frailty index (FI-lab) using routine blood tests.

Results: CFS, NEWS2 and FI-lab scores were independently associated with post-discharge survival in an adjusted Cox proportional hazards model with age, ward category (acute geriatric and general medical) and comorbidity as covariates. Adjusted hazard ratios and 95% confidence intervals were 1.54 (1.24-1.91) for CFS, 1.12 (1.03-1.23) for NEWS2, and 1.02 (1.00-1.05) for FI-lab. A frailty × illness severity category interaction effect (p = 0.003), suggested that the impact of frailty on survival was greater in those experiencing higher levels of illness severity. Among patients with at least moderate frailty (CFS six to eight) and high illness severity according to both NEWS2 and FI-lab, two (13%) were alive at follow-up.

Conclusion: Frailty screening aided prognostication of survival following discharge in older acutely ill persons admitted to medical wards. The prognostic value of frailty increased when combined with readily available illness severity markers acquired during admission.

Keywords: Acute care; Clinical Frailty Scale; Frailty index; Hospital complications; New Early Warning Score; Survival analysis.

Fig. 1

Flowchart of patient recruitment

Fig. 2

Kaplan–Meier curves illustrating estimated survival probabilities within follow-up according to different degrees of frailty. The survival curves according to CFS categories indicating no (1–3), very mild to mild (4–5), moderate (6) or severe to very severe (7–8) frailty were statistically different (Log-rank test, p < 0.001)

Fig. 3

a Kaplan–Meier curves display estimated survival probabilities within follow-up according to four categories of NEWS2 and FI-lab scores (high or low). Visual inspection suggests more favourable survival for patients with less severe illness. A log-rank test confirmed a significant difference between the survival curves. To visualize the interaction between NEWS2 and FI-lab categories and frailty, subjects were divided based a combined illness severity pattern emphasized by color: we considered low NEWS2 and FI-lab scores (green curve) as low; high NEWS2 (straight line) or FI-lab (dotted line) as intermediate (yellow curves); and high NEWS2 and FI-lab as high combined illness severity (red curve), respectively. b–d shows Kaplan–Meier plots of estimated survival probabilities for subjects living with no to mild (CFS 1–5; grey curves) or moderate to severe (CFS 6–8; black curves) frailty––stratified according to low, intermediate or high combined illness severity. Visual inspection of the curves indicates negligible differences in survival across categories of illness severity among patients living with no or mild frailty. The curves further suggests that the impact of living with moderate to severe frailty on survival probability is most pronounced in the face of more severe illness